Involvement of sphingosine kinase in plant cell signalling

Authors


*(fax +44 117 33179831; e-mail alistair.hetherington@bristol.ac.uk).

Summary

In mammalian cells sphingosine-1-phosphate (S1P) is a well-established messenger molecule that participates in a wide range of signalling pathways. The objective of the work reported here was to investigate the extent to which phosphorylated long-chain sphingoid bases, such as sphingosine-1-phosphate and phytosphingosine-1-phosphate (phytoS1P) are used in plant cell signalling. To do this, we manipulated Arabidopsis genes capable of metabolizing these messenger molecules. We show that Sphingosine kinase1 (SPHK1) encodes an enzyme that phosphorylates sphingosine, phytosphingosine and other sphingoid long-chain bases. The stomata of SPHK1-KD Arabidopsis plants were less sensitive, whereas the stomata of SPHK1-OE plants were more sensitive, than wild type to ABA. The rate of germination of SPHK1-KD was enhanced, whereas the converse was true for SPHK1-OE seed. Reducing expression of either the putative Arabidopsis S1P phosphatase (SPPASE) or the DPL1 gene, which encodes an enzyme with S1P lyase activity, individually, had no effect on guard-cell ABA signalling; however, stomatal responses to ABA in SPPASEDPL1 RNAi plants were compromised. Reducing the expression of DPL1 had no effect on germination; however, germination of SPPASE RNAi seeds was more sensitive to applied ABA. We also found evidence that expression of SPHK1 and SPPASE were coordinately regulated, and discuss how this might contribute to robustness in guard-cell signalling. In summary, our data establish SPHK1 as a component in two separate plant signalling systems, opening the possibility that phosphorylated long-chain sphingoid bases such as S1P and phytoS1P are ubiquitous messengers in plants.

Introduction

In animals, sphingosine-1-phosphate (S1P) has well-defined roles in cell signalling, especially in the control of growth and survival. It is synthesized from sphingosine by sphingosine kinase (SPHK), and is broken down by S1P phosphatase (SPPASE), the activity of which produces sphingosine or S1P lyase that degrades S1P to ethanolamine-1-phosphate and hexadecenal (Figure S1). A range of extracellular signals increase SPHK activity with consequent increases in intracellular S1P levels. There is evidence that once S1P is produced it can bring about direct intracellular responses, including an increase in cytosolic calcium concentration ([Ca2+]cyt). It can also be exported to elicit responses at the extracellular surface of the cell in which it was synthesized, or in other cells located nearby. It does this by binding to a family of receptors (S1P receptors) that are part of the G protein-coupled receptor (GPCR) superfamily (Chalfant and Spiegel, 2005; Ogretmen and Hannun, 2004; Spiegel and Milstien, 2003).

Investigations in our laboratory showed that plants contain S1P, that S1P levels increase during drought and that the application of S1P promotes stomatal closure in a Ca2+-dependent manner in Commelina communis. In order to investigate whether S1P was involved in guard-cell ABA signalling, we showed that dl-threo-dihydrosphingosine, which in mammalian cells acts as a competitive inhibitor of SPHK, partially attenuated ABA-induced stomatal closure (Ng et al., 2001). Working with Arabidopsis and using SPHK inhibitors, Coursol et al. (2003) confirmed that SPHK was involved in ABA-induced closure, and also in the ABA-mediated inhibition, of light-induced stomatal opening. They also demonstrated that guard-cell lysates exhibited SPHK activity, with sphingosine being the best substrate among the sphingoid bases assayed, and that guard-cell and mesophyll protoplasts rapidly and transiently formed S1P after treatment with ABA. In addition, they showed that S1P acts upstream of the sole prototypical putative heterotrimeric G protein α-subunit encoded by the GPA1 gene. More recently, Coursol et al. (2005) reported that the majority of Arabidopsis leaf SPHK activity is associated with membrane fractions, and that, in addition to sphingosine, SPHK is able to phosphorylate phytosphingosine, and that phytoS1P can, like S1P, induce stomatal closure.

By manipulating the expression and activity of SPHK in Arabidopsis we establish SPHK1 as a component in two different plant signalling networks, suggesting that phosphorylated long-chain sphingoid bases such as S1P and phytoS1P might be ubiquitous messengers involved in the control of other processes in plants. However, identifying the precise contribution of individual phosphorylated long-chain sphingoid bases to signalling must await subsequent analyses, and are not the subject of this current paper. We also observed that expression of SPHK1 and a putative SPPASE are coordinately regulated. This opens up the possibility that the homeostatic control of S1P or phytoS1P might be involved in regulating the robustness of stomatal function.

Results

Genes likely to encode enzymes of sphingosine-1-phosphate metabolism

We identified and characterized candidate Arabidopsis SPHK and SPPASE genes. We also studied the recently characterized DPL1 gene, which is a long-chain base phosphate lyase that uses S1P as a substrate (Tsegaye et al., 2007). Inspection of the Arabidopsis genome revealed four candidate genes showing homology to human and mouse SPHK genes. Of these four candidate genes, one (At5g51290) encodes a ceramide kinase (AtCERK) (Liang et al., 2003), whereas a second (At5g23450) encodes a long-chain base kinase, designated as AtLCBK1 (Nishiura et al., 2000). We have now investigated the two remaining genes that are likely to encode SPHK: At4g21540 and At2g46090.

Figure S2 shows a partial alignment of the two putative Arabidopsis SPHK and AtLCBK1 sequences across the five conserved domains that are characteristic of sphingosine kinases, together with deduced amino acid sequences from the two human SPHK genes and the yeast LCB kinase 4. Only At4g21540 possesses the conserved aspartate residue in conserved domain C4 believed to be involved in sphingosine binding (Yokota et al., 2004). The protein encoded by At4g21540 shows the greatest percentage identity with the well-characterized human SPHK enzymes. In terms of amino acid sequence identity, At4g21540 shows a slightly higher (30%) identity with human SPHK1 than human SPHK2 (27%).

Inspection of the Arabidopsis genome revealed a good candidate open reading frame that was likely to encode an SPPASE (At3g58490). Figure S3 shows a partial amino acid sequence alignment with amino acid sequences derived from genes of other organisms. When compared with SPPASEs from mammals and yeast, the predicted Arabidopsis protein sequence shares 37–44% similarity and 20–26% identity. More significantly, the Arabidopsis sequence contains the three conserved domains that are predicted to be involved with hydrolysis of the phosphate moiety in lipid phosphohydrolase and SPPASE (Le Stunff et al., 2002).

Substrate specificity of enzymes encoded by sphingosine kinase genes

In order to investigate whether At4g21540 and At2g46090 encode enzymes capable of producing S1P, we expressed these coding regions in human embryonic kidney 293 (HEK 293) cells. Figure 1 shows that lysates from HEK cells expressing At4g21540 phosphorylated sphingosine. Accordingly, we designated this gene product as SPHK1. We decided upon SPHK1 (as opposed to, for example, LCBK2) because this designation fits well with previous guard-cell work on sphingosine kinase activity (Coursol et al., 2003, 2005), and reflects the established role of S1P in guard-cell signalling. SPHK1 exhibited much higher activity when its substrate was presented as a complex with BSA, rather than in Triton X-100 micelles (Figure 1a,b), and its activity was markedly stimulated by KCl (Figure 1c). SPHK1 is also capable of using other plant long-chain sphingoid bases, including phytosphingosine, 4-hydroxy-8-sphingenine (t18:1), 4,8-sphingadienine (d18:2) and dihydrosphingosine (sphinganine) (Figure 1d). Similar to its human homologs, SPHK1 did not phosphorylate N,N-dimethylsphingosine. Additionally, SPHK1 did not phosphorylate C2 ceramide (data not shown). Using a similar assay, lysates from cells expressing At2g46090 did not show significant sphingosine phosphorylating activity above that of the vector control, despite being expressed to a similar level as SPHK1 (data not shown), suggesting that it may not be a bona fide sphingoid base kinase (nevertheless, we cannot formally exclude the possibility that our assay conditions are inappropriate for detecting the enzyme activity of the At2g46090 gene product). Therefore, we concentrated our efforts on characterizing and investigating the role of the SPHK1 gene product.

Figure 1.

 Phosphorylation activity of putative Arabidopsis sphingosine kinases (SPHKs). Human embryonic kidney 293 (HEK 293) cells were transfected with the indicated plasmids, and phosphorylation activity was measured in cell lysates with sphingosine added with Triton X-100 (a) or as a complex with BSA (b). HEK 293 cells were transfected with plasmid containing At4g21540, and phosphorylation activity was measured in cell lysates with sphingolipid added under the conditions indicated ± 1 m KCl (c). HEK 293 cells were transfected with plasmid containing At4g21540 (d). Cell lysate proteins were separated by SDS-PAGE and were immunoblotted with anti-V5 antibody to show expression of the recombinant proteins (inset). Data are means ± SE of three independent experiments, each performed in duplicate. hSPHK1: human sphingosine kinase 1 (accession number AAF73423).

Manipulating the expression of SPHK1

Figure 2(a) indicates the position of T-DNA insertions in SPHK1 in two independent Arabidopsis lines. Relative to wild type, the GABI KAT line overexpressed the SPHK1 transcript, whereas the SAIL T-DNA line exhibited a reduction in SPHK1 transcript abundance (Figure 2b). Hereafter, these T-DNA lines are referred to as SPHK1-OE and SPHK1-KD. Measurements of the enzyme activity in different subcellular fractions indicated that the majority of SPHK activity was located in the membrane fraction of wild-type plants, and that most of this activity was abolished in the SPHK1-KD line, whereas it was increased in the SPHK1-OE line (Figure 2c).

Figure 2.

 Characterization of SPHK1 T-DNA insertion Arabidopsis lines.
(a) Diagram showing the positions of T-DNAs in the two independent insertion lines. Quantitative RT-PCR analysis of SPHK1 gene transcript levels in the T-DNA insertion lines compared with Col-0 control (b). Phosphorylation activity in extracts from different subcellular fractions of Col-0, SPHK1-OE and SPHK1-KD Arabidopsis plants using 50 μm sphingosine complexed with BSA as the substrate in the presence of 1 m KCl. Data are means ± SD of two independent experiments, each performed in duplicate (c).

Involvement of SPHK1 in guard-cell ABA signalling

Figure 3(a,b) shows that stomata of SPHK1-KD are less sensitive, and that stomata of SPHK1-OE are more sensitive, than wild type to the application of ABA. Similarly, when these two lines were examined for the ability of ABA to inhibit light-induced stomatal opening, the KD line was less sensitive and the OE line was more sensitive to applied ABA than the wild type (Figure 3c). These results firmly establish that SPHK1 is involved in guard-cell ABA signalling.

Figure 3.

 SPHK1 has a role in two ABA-signalling pathways. Comparison of SPHK1 T-DNA insertion lines with Col-0 in an ABA-induced promotion of stomatal closure assay (a) and (b). Epidermal strips were incubated under opening conditions, and then stomatal apertures were measured after subsequent treatment with or without ABA for 2.5 h. Values are means ± SE: (a) = 120; (b) = 80. Comparison of SPHK1 T-DNA insertion lines with Col-0 in an ABA inhibition of stomatal opening assay (c). Stomatal apertures were measured after 3 h of incubation with or without ABA. Values are means ± SE (= 120). Statistical analyses were performed by Student’s t-test: **< 0.001. The experiments were repeated three times, except for 3C, which was repeated twice. Germination of SPHK1 T-DNA insertion lines compared with Col-0 (d). The experiment was repeated three times. Percentage radicle emergence 1 day after imbibition: the probability of germination was significantly different between lines [generalized linear model (GLM), deviance = 87.86, degrees of freedom (df) = 2, P < 0.0001] and treatments (GLM, deviance = 25.36, df = 2, P < 0.0001).

Involvement of SPHK1 in the regulation of seed germination

Figure 3(d) shows that the germination of mature seed was delayed in the SPHK1-OE seed, and was more rapid in the SPHK1-KD seed, in comparison with wild type. Seed viability was not affected, as all mutant and control lines reached 100% germination on extended incubation (data not shown). When we germinated SPHK1 mutant seed on exogenous ABA, we found that, in line with the situation in guard cells, the germination of SPHK1-OE was more sensitive, and the germination of SPHK1-KD was less sensitive, than wild type to applied ABA.

Downregulating the expression of putative SPPASE and DPL1 genes

In order to investigate whether the putative SPPASE and DPL1 genes are involved in ABA signalling, we used RNAi to downregulate their expression. Figure 4(a) shows that in two separate RNAi lines, DPL1 transcript abundance was reduced significantly relative to wild type (set at unity). This was also the case for two separate SPPASE RNAi lines (Figure 4b), and for two independent double SPPASE RNAi DPL1 RNAi lines (independent crosses, based on four lines) (Figure 4c).

Figure 4.

 RNAi silencing of SPPASE and DPL1 in Arabidopsis plants. Quantitative RT-PCR analysis of SPPASE and DPL1 transcript levels in single (a and b) and double SPPASE DPL1 RNAi transgenic plants (c). The experiment was repeated three times.

Characterizing the phenotypes of the SPPASE and DPL1 RNAi plants

We investigated whether the SPPASE and DPL1 RNAi plant lines were compromised in guard-cell ABA signalling. Neither showed significant differences to wild type in either the ABA inhibition of light-induced stomatal opening or the ABA promotion of stomatal closure (data not shown). However, we found that the double SPPASEDPL1 RNAi stomata were more sensitive to applied ABA in the inhibition-of-opening bioassay (Figure 5a). In contrast, both independent double RNAi lines closed in response to applied ABA in a manner that was indistinguishable from the wild type (Figure 5b). We also investigated the germination of the putative SPPASE and DPL1 RNAi lines. Whereas the DPL1 RNAi seed showed no significant difference to wild type, the two SPPASE RNAi lines exhibited increased sensitivity to ABA during germination in comparison with wild type (Figure 5c).

Figure 5.

 Characterization of SPPASEDPL1 RNAi plants. Comparison of double mutant SPPASE DPL1 RNAi lines with Col-0 in an inhibition of stomatal opening assay. Apertures were measured after 3 h of incubation with or without ABA (a). Comparison of double mutant SPPASE DPL1 RNAi lines with Col-0 in an ABA-induced promotion of stomatal closure assay. Epidermal strips were pre-treated under opening conditions, and then stomatal apertures were measured after subsequent treatment ± ABA for 2.5 h (b). Values are means ± SE (n = 120). Statistical analyses were performed by Student’s t-test: *P < 0.05; **P < 0.001. Germination of single RNAi lines compared with Col-0 (c). The experiment was repeated three times. The percentage radicle emergence 3 days after imbibition: the probability of germination was significantly different between lines [generalized linear model (GLM), deviance = 100.1, degrees of freedom (df) = 4, P < 0.0001] and treatments (GLM, deviance = 198.8, df = 2; P < 0.0001). There was a significant interaction between line and treatment (GLM, deviance = 25.94, df = 8, P < 0.005), because the treatment had different effects depending on which line it acted upon; the germination reduction that is caused by ABA is greater in SPPASE RNAi-3 and SPPASE RNAi-4 than in the other lines.

Coordinated regulation of SPHK1 and putative SPPASE gene expression

To understand the possible homeostatic control of S1P and phytoS1P levels, we investigated whether the putative SPPASE and DPL1 mRNA transcript abundances might be altered in the SPHK1-OE and SPHK1-KD backgrounds. Figure 6 shows that SPPASE transcript abundance is upregulated in the SPHK1-OE background and downregulated in the SPHK1-KD background, whereas DPL1 transcript abundance is not significantly different to that found in the wild type. These results suggest that there is coordinated feedback regulation of SPHK1 and SPPASE gene expression that might be part of an S1P/phytoS1P homeostatic mechanism that could contribute to robustness in guard-cell signalling.

Figure 6.

 Altered SPPASE expression in SPHK1 T-DNA insertion lines.
Quantitative RT-PCR analysis of SPHK1, SPPASE and DPL1 transcript levels in SPHK1-OE and SPHK1-KD plants. The experiment was repeated three times.

Discussion

Characterizing the SPHK and SPPASE genes

In this investigation we showed that SPHK1 phosphorylates sphingosine and other long-chain sphingoid bases, including phytosphingosine (Figure 1). In these respects, the activity of SPHK1 is more like the SPHK activity described from Arabidopsis leaves (Coursol et al., 2005) than the product of the long-chain base kinase gene (LCBK1), which prefers d-erythro dihydrosphingosine to sphingosine. Furthermore, LCBK1, unlike SPHK1, will not phosphorylate dl-threo dihydrosphingosine (Imai and Nishiura, 2005). In addition, and again like the SPHK from Arabidopsis leaves (Coursol et al., 2005), there was greater SPHK1 activity associated with membranes than in a soluble preparation (Figure 2), in contrast with the situation of mammalian SPHK1, where greater activity is reported in the soluble fraction (Taha et al., 2006). Interestingly, SPHK1 exhibits some properties that are similar to murine SPHK1, namely an ability to phosphorylate d-erythro dihydrosphingosine, whereas other features are more similar to murine SPHK2, including the ability to phosphorylate a range of substrates, including dl-threo dihydrosphingosine, phytosphingosine, stimulation by KCl and inhibition by Triton X-100 (Taha et al., 2006). In this respect, SPHK1 is similar to the SPHK activity investigated in Arabidopsis leaves, which also preferred its substrates presented in a complex with BSA, as opposed to in the presence of Triton X-100 (Coursol et al., 2005).

The involvement of SPHK1, SPPASE and DPL1 genes in guard-cell ABA signalling

We identified SPHK1-KD and SPHK1-OE plant lines that were characterized by decreased or increased SPHK1 transcript levels and enzyme activity, respectively (Figure 2). When the stomata of these lines were tested for their ability to close in response to ABA, compared with the wild type, it was apparent that the KD line was less sensitive, and that the OE line was slightly more sensitive, to applied ABA (Figure 3). These data show that the SPHK1 gene product is involved in guard-cell ABA signalling.

In response to the addition of ABA, stomatal behaviour in SPPASE RNAi and DPL1 RNAi plant lines was unaffected. However, the inhibition of stomatal opening by ABA was partially compromised in the double SPPASEDPLI RNAi plants (Figure 5). At this stage we cannot rule out the possibility that other mechanisms exist that are able to compensate for the loss of these gene products, or that SPPASE and DPL1 gene expression was not completely silenced by the RNAi. The former possibility is borne out to some extent by the presence of: (i) the At5g03080 gene, which although exhibiting low sequence similarity with other SPPASE genes, does possess the three conserved domains that are characteristic of phosphatases; and (ii) open reading frames within the Arabidopsis genome that are likely to include general lipid phosphate phosphohydrolases. In animal cells, these general phosphohydrolases are able to use S1P as a substrate (Imai and Nishiura, 2005). Our data suggest that the SPPASE and DPL1 are involved in the ABA inhibition of the stomatal opening pathway, but are not involved in the pathway by which ABA promotes stomatal closure. These data fit well with other observations suggesting that ABA controls stomatal aperture by a pathway that bifurcates in such a way that one wing brings about promotion of closure, and the other brings about the inhibition of opening (Mishra et al., 2006).

Robustness in the guard-cell signalling system

We have noted in the past that stomatal function is robust towards perturbation (Hetherington and Woodward, 2003). Our present data have a bearing on one of the possible bases of robustness, as compensation for the loss or partial loss of SPHK activity could be gained through an increase in the expression (or activity) of another member of the SPHK gene family, or of another gene involved in S1P metabolism. As we could find no evidence that LCBK1 expression was altered in either the SPHK1-OE or SPHK1-KD backgrounds (data not shown), we measured SPPASE transcript abundance in the SPHK1-KD and SPHK1-OE backgrounds. This revealed that SPPASE expression was increased or decreased, respectively, suggesting that transcripts of SPHK1 and SPPASE are controlled in a coordinated manner (Figure 6). If these changes in transcript abundance are reflected in enzyme activity, then the overall effect of controlling the activity of the kinase–phosphatase pair might be to maintain control of S1P or phytoS1P levels in the face of perturbation of the kinase. It is noteworthy in this respect that there was no evidence that DPL1 transcript abundance was coordinately regulated. This latter result might be readily explained in the homeostatic context, as the SPHK1–SPPASE pair both control S1P/phytoS1P homeostasis, whereas the presumed products of DPL1 activity, phosphoethanolamine and hexadecanal, are unlikely to contribute to S1P/phytoS1P homeostasis directly. However, further work involving measuring S1P levels and the SPPASE activity is required before more definitive statements can be made about the possible mechanisms underlying robustness. In addition, although we saw no evidence of coordinated regulation of the LCBK1 gene in the SPHK1-OE and SPHK1-KD backgrounds, it would be interesting to investigate the extent to which guard-cell signalling is compromised in a lcbk1 sphk1 double mutant.

The involvement of SPHK1, SPPASE and DPL1 genes in germination

The results shown in Figure 3(d) show that germination is significantly affected in SPHK lines. When we investigated the ability of applied ABA to inhibit germination, we found that SPHK-KD is more able, and SPHK-OE is less able, to germinate in the presence of exogenously supplied ABA, and this was mirrored in the SPPASE RNAi seed, which showed an increased sensitivity to ABA during germination (the DPL1 RNAi seed exhibited wild-type behaviour). The increased rate of germination, and enhanced ability to germinate on exogenous ABA, phenotypes of our SPHK1-KD mutant are reminiscent of other Arabidopsis mutants that have been previously described as ABA-deficient or ABA-insensitive (Koorneef et al., 1984). A subset of these mutants, abi1 and abi2, also show increased transpiration and/or reduced ABA inhibition of stomatal opening (Roelfsema and Prins, 1995). Our data are consistent with a role for SPHK1 in the control of germination by ABA. Interestingly, on the basis of the results presented here, it would seem that the knock-down of the SPHK1 gene has a more profound effect on germination than on ABA-mediated stomatal movement. We have already suggested (above) that the robust nature of guard-cell signalling, and specifically compensation in the system, might be responsible for the smaller effect seen in stomata. However, another possible explanation would be that there is a tissue-specific element in signalling systems with, in this case the S1P system assuming greater importance in seed than in guard cells. This possibility awaits further investigation.

The role of SPHK1 in plant cell signalling

In summary, our data show that SPHK1 activity is required in ABA-induced stomatal movements and in the control of germination. How might SPHK1 function in plant cell signalling? From the results reported here and in previous papers, we know that SPHK1 will phophorylate sphingosine, phytosphingosine and, to a lesser extent, other long-chain sphingoid bases. We also know that the addition of S1P to guard cells results in an elevation of cytosolic Ca2+, and to a reduction in stomatal aperture. In addition, although we do not know whether phytoS1P is capable of generating increases in cytosolic Ca2+, we do know that it will bring about stomatal closure (Coursol et al., 2003, 2005; Ng et al., 2001). In this study, we have neither addressed the issues of site(s) of S1P/phosphorylated long-chain sphingoid base synthesis nor attempted to quantify any of the species of the phosphorylated long-chain sphingoid base. These investigations will be the subject of future experiments. However, based on the activity observed in the leaf extracts reported here, and from the activity reported from guard and mesophyll cell protoplast data (Coursol et al., 2005), it would seem likely that S1P, at least, is synthesized in both its target cells (then exported) and elsewhere. In either case an export/transport system will also be required. Currently, neither the receptor nor the transporter(s) have been identified. Clearly further work is required to resolve the mechanisms of action of S1P (and phytoS1P) in plants. However, the data reported here provide conclusive evidence that SPHK1 is involved in guard-cell ABA signalling, and in the control of germination. Taken together, these results suggest that S1P and phytoS1P are likely to be ubiquitous messengers involved in plant cell signalling.

Experimental procedures

Plant material

Arabidopsis (ecotype Col-0) seed germinated on agar (1/2 strength MS, 1% sucrose and 0.6% agar) and transferred to peat-based compost:sand mix (3:1) at 10-days old, were grown at 200 ± 20 μmol m−2 sec−1 (Osram Powerstar HQI-BT 400W/D; Osram, http://www.osram.com) with a 10-h photoperiod at 22 ± 2°C, and with a relative humidity of 55 ± 5%.

Cloning the SPHK1At5g21540 transcript

The Arabidopsis genome contains a tandem repeat of apparent SPHK encoding units in the region of the At4g21540 locus, only the second of which is currently annotated as transcribed. This annotation appears to be correct, as RT-PCR experiments in this study only successfully amplified cDNA from the second SPHK encoding repeat; but, we note that Tsegaye et al. (2007) report, as unpublished observations, that a transcript from At4g21535 is expressed in senescent tissues. In our study, RNA was prepared from 6-week-old Arabidopsis plants following 24 h of drought, as described by Verwoerd et al. (1989), and 5 μg of RNA was reverse transcribed with oligonucleotide primer OG1 (primer sequences are given in Table S1). One fiftieth of the cDNA was used as a PCR template, with primers SPHK-5′ and 1996f designed to amplify the first repeat, and primers 5389f, 5581r and SPHK-3′ designed to amplify the second repeat for 35 thermocycles. The 5389f/SPHK-3′ and 5389f/OG1 primer combinations yielded 1.4- and 2-kb fragments, respectively, which were cloned and sequenced. Primers designed to amplify across both repeats did not generate any products. A 5′ RACE kit (Gibco-BRL) was used to determine the 5′ end of the SPHK1 transcript. Primer 5648r was annealed to 5 μg RNA and first-strand cDNA was synthesized, which was then purified and TdT tailed. cDNA was used in a PCR with nested primer 5581r and kit anchor primer. The ∼400-bp PCR products were cloned and sequenced. Six clones were identical except for the 5′ ends, which extended to different extents up to 115 bp 5′ of the 5389f/SPHK-3′ RT-PCR product. The full SPHK1 coding region encoding 498 amino acids was then amplified from first-strand cDNA with primers per8f and per8r, and was then cloned and sequenced (Geneservice DNA Sequencing Facility, http://www.geneservice.co.uk). The deduced amino acid sequence from cloned SPHK1 is identical to that deduced from the full-length RAFL clone RAFL08-18-M20.

Construction of mammalian expression plasmids

The SPHK gene coding regions were amplified from plasmids with KOD polymerase (Novagen, http://www.merckbiosciences.co.uk/html/NVG/home.html) and per8f, 1V5HIS-r, per8r, SPHK2-5′, Kozak SPHK2 and 2V5HISr primers. PCR products were cloned directionally into pcDNA3.1/V5-His plasmid vector (Invitrogen, http://www.invitrogen.com) and were sequenced.

Cell culture and transfection

The HEK 293 cells (ATCC CRL-1573) were cultured in high glucose Dulbecco’s modified Eagle’s medium supplemented with 10% fetal bovine serum. Cells were cultured on poly-d-lysine-coated plates, and were then transfected with Lipofectamine Plus (Invitrogen). The transfection efficiency was typically 95%.

SPHK activity assay

Sphingosine, dihydrosphingosine, dl-threo dihydrosphingosine, N,N′-dimethylsphingosine and phytosphingosine were obtained from Biomol Research Laboratory (http://www.biomol.com). Both t18:1 and d18:2 were a kind gift from Dr D.V. Lynch (Williams College, Williamstown, MA, USA). Stock solutions were prepared in ethanol (Sph, dhSph and PhytoSph) or methanol (S1P, dhS1P, DMS, d18:2 and t18:1). Prior to use, aliquots of lipids were dried under a stream of nitrogen, and were solubilized either in 5% Triton X-100 or 4 mg ml−1 fatty acid-free BSA (Sigma-Aldrich, http://www.sigmaaldrich.com). The HEK 293 cells were lysed by freeze-thawing in 20 mm Tris (pH 7.4), 20% glycerol, 1 mm 2-mercaptoethanol, 1 mm EDTA, 5 mm sodium orthovanadate, 40 mmβ-glycerophosphate, 15 mm NaF, 0.5 mm 4-deoxypyridoxine and Sigma protease inhibitor cocktail. Unbroken cells were removed by centrifugation at 1000 g for 10 min. SPHK activity was measured essentially as previously described (Olivera and Spiegel, 1998). Briefly, cell extracts (10 μg) were incubated at 37°C with 10 μCi, 1 mm [γ-32P]ATP (PerkinElmer, http://www.perkinelmer.com) in 10 mm MgCl2 and 50 μm sphingolipid substrate, added either in micellar form with Triton X-100 (final concentration 0.25%) or as a BSA complex (final concentration 0.2 mg ml−1). Reactions were stopped with 800 μl chloroform:methanol:concentrated HCl (100:200:1 v/v/v), 250 μl chloroform and 250 μl 2 m KCl. Labelled lipids were extracted and separated by TLC on silica gel G60 with chloroform:acetone:methanol:acetic acid:H2O (10:4:3:2:1, v/v) as the solvent. Radioactive bands were quantified with an FX Molecular Imager (Bio-Rad, http://www.bio-rad.com). SPHK specific activity is expressed as pmol phosphorylated lipid formed min−1 mg−1 protein.

Characterization of SPHK1 T-DNA insertion plants

GABI-KAT (Rosso et al., 2003) line 288 D07 and SAIL (Sessions et al., 2002) line 794 B1 seeds were collected from individual plants. Resistant seedlings were selected by sowing line 288 D07 on 1/2 strength MS plates containing 5.25 mg L−1 sulfadiazine (Sigma-Aldrich), or by spraying line 794 B1 with glufosinate (Liberty; Aventis, http://www.sanofi-aventis.co.uk), 20 ml (1 mm solution) sprayed over approximately 30 plants at the five-leaf stage. Seedlings of both lines showed an approximately 3:1 resistant:sensitive segregation ratio. Homozygous plants were obtained for both lines and the seed was bulked up. The position of the T-DNA insertion within the At4g21540 gene was confirmed in each line by sequencing of T-DNA border PCR products. Line 794 B1 was homozygous for a single T-DNA locus present within the SPHK1 gene, which was confirmed by southern blotting.

Leaf protein extraction

Six-week-old Arabidopsis plant leaves were homogenized directly in SPHK buffer (Coursol et al., 2005). The homogenate was centrifuged at 10 000 g for 10 min. Cytosolic and membrane fractions were obtained from the resulting supernatent by centrifugation at 100 000 g for 30 min. Samples were frozen in liquid N2 and stored at −80°C.

Production of SPPASE and DPL1 RNAi-expressing Arabidopsis plants

Partial cDNA fragments of SPPASE and DPL1 were amplified using the gene-specific primer sequences (see Table S1) with KOD hot-start DNA polymerase (Novagen), cloned into the pENTR/D-TOPO® (Invitrogen) and were then transferred to the pK7GWIWG2(I) destination vector (Karimi et al., 2005). After sequence verification, constructs were transformed into Arabidopsis plants using the Agrobacterium-mediated floral-dip method (Clough and Bent, 1998). T2 plants that displayed 3:1 (kanamycin resistant:sensitive) segregation were self fertilized to obtain seed stocks. T3 or T4 generation plants were used for our investigations. The DPL1SPPASE RNAi plants were obtained by performing crosses between homozygous DPL1 RNAi and SPPASE RNAi plants. The resulting F1 plants were used in our studies.

Quantitative RT-PCR analysis

RNA was extracted from 20-day-old seedlings and cDNA was synthesized as above except with 1 μg of total RNA as the template. RT-qPCR was performed in an ABI Prism 7000 cycler (Applied Biosystems, http://www.appliedbiosystems.com) using the TaqMan® Universal PCR Master Mix, cDNA corresponding to 20 ng of total RNA, 250 nm of each primer and 100 nm of probe in a 25-μl reaction at 50°C for 2 min, and 95°C for 10 min, followed by 40 two-step cycles at 95°C for 15 sec and 60°C for 1 min. Gene-specific primers and TaqMan® MGB probes were designed with Primer Express 2.0 software (Applied Biosystems). See Table S1 for primer/probe details. The relative RNA levels were calculated from cycle threshold (CT) values according to the ΔCT method (Applied Biosystems), and relative SPHK1, DPL1 and SPPASE mRNA levels were normalized to Actin3 mRNA levels. Reactions were repeated independently three times.

Stomatal aperture bioassay

Inhibition of stomatal opening by ABA in 5–6-week-old plants was investigated as described previously, except that incubation was at 22°C (Webb and Hetherington, 1997). The ABA-induced promotion of stomatal closure on isolated epidermal strips was conducted in the same way, except that the epidermal peels were incubated for 2.5 h under stomatal-opening conditions, ABA was added to the incubation buffer and, after a further 2.5-h incubation, stomatal apertures were measured. The experiments were repeated at least three times with similar results, except for the inhibition of opening assay (Figure 3c), which was repeated twice with similar results. Aperture measurements from individual experiments were pooled, and significance was assessed using the Students t-test.

Germination assay

Seeds used in each experiment were collected from plants grown and stored together. Mature dry seeds were surface sterilized in 70% (v/v) ethanol, and were sown in lots of 52, or 25 in the case of the RNAi lines, in three replicates, in 9-cm-diameter Petri dishes containing 30 ml double distilled H2O solidified with 8 g L−1 agar (Sigma-Aldrich), supplemented with ABA where appropriate. Seeds were germinated under a 16-h photoperiod with 150 μmol m−2 sec−1 (Philips MASTER TLD 36W/840 tubes; Philips, http://www.philips.com) at 22°C (light period)/20°C (dark period), and were then scored for radicle emergence each day for 7 days. The data were analysed using a generalized linear model (GLM) with binomial errors and a logit link using S-Plus 8.0 (Venables and Ripley, 1999). The starting model, full factorial with three factors (treatment, line and replicate), was simplified by the removal of non-significant interactions and terms following the method used by Crawley (1993). Experiments were repeated three times with similar results.

Acknowledgements

DW, YKL, GHH, JEG and AMH acknowledge funding from the UK BBSRC. YKL and AMH also acknowledge support from the Gatsby Charitable Foundation and the University of Bristol. SS acknowledges support from NIH R37GM043880. The authors are grateful to Dr Ian Hartley (Lancaster University) for help with statistical analyses and Dr Daniel Lynch (Williams College, Williamstown, MA, USA) for the gift of 4-hydroxy-8-sphingenine (t18:1) and 4, 8-sphingadienine (d18:2).

Ancillary